LRTI represent a major global health burden due to their substantial morbidity and significant mortality rates¹³. The etiological complexity of LRTIs is underscored by the diverse spectrum of potential pathogens, encompassing bacteria, viruses, fungi, and atypical microorganisms, all of which may contribute to disease progression under specific clinical conditions. Current conventional diagnostic modalities for LRTIs—including culture-based methods and targeted molecular assays—are constrained by limited sensitivity, prolonged turnaround times, and narrow pathogen coverage, often resulting in delayed or suboptimal antimicrobial therapy¹⁴. In contrast, mNGS offers advantages in pathogen detection by providing an unbiased, hypothesis-free analysis of microbial communities within clinical specimens. This capability enables comprehensive identification of causative pathogens, including fastidious or co-infecting organisms¹⁵. BALF serves as a suitable specimen for mNGS-based pathogen detection in LRTIs. To evaluate the diagnostic utility of this approach, our retrospective study performed a comparative analysis of mNGS and culture using BALF samples from suspected LRTI patients. Although preliminary evidence supports the enhanced sensitivity of mNGS, there is some unexplained heterogeneity in the diagnostic efficacy of mNGS in the study.

We conducted a retrospective cohort study to assess the diagnostic utility of mNGS and culture methods in 400 patients with suspected LRTI. Of them, 329 patients were identified with LRTI, 71 were ruled out of LRTI. mNGS provides a more excellent coverage of pathogen detection than culture methods. The study observed that 93.3% of those diagnosed with LRTIs obtained pathogenic microorganism results through mNGS, the culture methods identified infections in only 55.6% of cases. The most commonly detected bacteria included Acinetobacter baumannii, Klebsiella pneumoniae, Mycobacterium tuberculosis, and Pseudomonas aeruginosa, conforming to typical bacteria prevalent in LRTI^16,17. In addition, mNGS identified atypical pathogens in cases where culture yielded negative results, for example, Chlamydia psittaci, Pneumocystis jirovecii, Leptospira interrogans, Nontuberculosis mycobacteria and Mycobacterium Tuberculosis. This expanded diagnostic capability positions mNGS as a critical tool for guiding targeted antimicrobial therapy¹⁸. When mNGS and culture are compared, it is found that there is a remarkable lack of consistency in the detection rates of pathogenic microorganisms. 45.0% (180/400) of the patients got positive results at the same time. of the 180 double-positive patients, 27 cases (6.8%) were partially matched. mNGS identify pathogens more comprehensively in mixed pulmonary infection. This indicates that mNGS can detect pathogenic microorganisms not identified by culture methods¹⁹. While our study demonstrates the superior sensitivity of mNGS in pathogen detection, a critical analysis revealed that 3.6% (12/329) of LRTI patients had pathogenic bacteria confirmed by culture but not detected by mNGS. False negatives detected by mNGS primarily comprised bacteria that can be easily cultured, including Acinetobacter baumannii(n = 3), Candida albicans(n = 3), Klebsiella pneumoniae(n = 2), and Pseudomonas aeruginosa(n = 2), Salmonella(n = 1), and Candida krusei(n = 1).These findings aligned with previous studies²⁰. On the one hand, Acinetobacter baumannii, Klebsiella pneumoniae, and Pseudomonas aeruginosa are commonly colonized in the respiratory tract, and the stringent threshold settings of mNGS may be filtered during bioinformatic analysis²¹. On the other hand, inadequate sequencing depth or filtering parameter settings could also lead to the failure in identifying low-abundance pathogens²².

mNGS demonstrates higher sensitivity and lower specificity in identifying the etiology of LRTI patients compared to culture, consistent with a previous research of 240 patients with suspected pulmonary infection²³. Based on the unbiased and highly sensitive characteristics of mNGS, failure to correctly interpret the test report may lead to erroneous positive results²⁴. Our study detected additional sequences in 71 patients who were ultimately diagnosed with non-LRTIs. Based on the patient’s medical history and clinical symptoms, most viruses are unrelated to clinical infections. There is still no uniform standard to determine whether the virus detected in the sample is pathogenic or only present as a carrier²⁵. EBV and CMV can establish lifelong latent infection. When the carrier’s immune function is suppressed, it can be reactivated and cause infection²⁶. The clinical judgment of the virus was made by several experienced clinicians based on the patient’s medical history and clinical symptoms. Other false positives likely result from contamination in sample collection, colonizing bacteria, disrupted mucosal barrier function, and library preparation^27,28. It is a challenge for clinicians to distinguish pathogens from colonizing bacteria and environmental pollutants.

Determining the responsible pathogen is the key to transforming empirical treatment into targeted treatment. When the diagnosis is unclear, clinical doctors will empirically apply broad-spectrum antibiotics, which may exacerbate the emergence of multi-drug-resistant pathogens²⁹. mNGS brings unique opportunities and challenges for providing personalized antimicrobial drug management decisions³⁰. Early diagnosis of mNGS can achieve early targeted treatment, shorten patient hospitalization time and reduce hospitalization costs³¹. However, studies have differing views about mNGS’ impact on treatment strategies, which can be attributed to differences in patient populations and study designs. Liang et al. reported that over 80% of patients adjusted or confirmed antibiotic treatment based on mNGS³². In contrast, Niles et al. showed that in most cases, little or no additional value was gained when mNGS was ordered concurrently with routine testing³³. 50.5% altered their antibiotic regimen based on pathogenic microbiological results of mNGS in our study, including escalating the treatment by increasing dosage or medication, de-escalating the empirical antibiotic treatment, or adjusting antimicrobial regimens. A prospective observational study on the clinical utility of mNGS for LRTIs in clinical outcomes found that patients with LRTIs showed significant improvement in clinical symptoms and inflammatory indicators after adjusting treatment according to mNGS results⁸. mNGS results offer an opportunity for targeted therapy, which may improve clinical antibiotic overuse to reduce resistance and medical resource waste.

mNGS also has some significant flaws that cannot be ignored. First, the high sensitivity of mNGS means that microbial contamination in the environment or reagents and colonizing microorganisms in the human skin will affect the large number of non-pathogenic organisms contained in the report, and the pathogens are hidden in the colonization and background microorganisms(false positive problem)³⁴. Second, for intracellular bacteria and some pathogens with rigid cell walls, such as fungi, the extraction efficiency of nucleic acid is low, leading to low detection sensitivity of both(false negative problem)³. Third, the cost of mNGS is relatively high³⁵.

Our study has some limitations. First, its retrospective design may introduce selection bias, as only patients with suspected LRTIs undergoing BALF sampling were included. Second, the lack of universally standardized criteria for distinguishing true pathogens from colonization or environmental contaminants in mNGS results necessitates cautious interpretation. Third, our single-center design and reliance on multidisciplinary clinical adjudication (rather than a predefined diagnostic protocol) may affect the generalizability of findings. Future studies should adopt multicenter prospective designs, and establish consensus guidelines for mNGS interpretation to optimize its diagnostic accuracy and clinical utility.